Thoracic Ossification of the Posterior Longitudinal Ligament: Thickness Changes After Posterior Decompression and Corrective Fusion in a Longitudinal Study With up to 10-Year Follow-up

Abstract

Study Design

Retrospective cohort study.

Objectives

Thoracic ossification of the posterior longitudinal ligament (OPLL) is traditionally considered progressive, raising concern that the ossified mass left after posterior decompression and instrumented fusion (PDF) may enlarge. We characterized long-term CT-based changes in focal OPLL thickness after PDF and assessed whether the residual mass continued to enlarge.

Methods

We reviewed all consecutive adult patients who underwent PDF for thoracic OPLL (2009–2023) with ≥2 years of follow-up. Maximum OPLL thickness was measured on axial CT and tracked longitudinally; a predefined 2-mm threshold, prespecified as the minimal clinically important difference, classified change. Thoracic kyphosis (TK) and modified Japanese Orthopaedic Association (mJOA) scores were also assessed.

Results

Twenty-six patients (mean age 49.5 ± 12.4 years; 15 men) were followed 5.4 ± 2.7 years (range 2–11). Mean axial OPLL thickness changed from 7.02 ± 1.58 to 6.90 ± 2.01 mm (Δ −0.12 ± 1.03 mm). By the 2-mm threshold, 92.3% were stable, 3.8% decreased, and 3.8% increased; directional decreases occurred in 69.2% and increased with longer follow-up (37.5%, 77.8%, 88.9%; p = 0.023). TK increased (+6.2°, p < 0.001) but was uncorrelated with thickness change (r = 0.250). mJOA improved from 7.3 to 8.7 (p = 0.0001; recovery rate 34.9%; excellent/good 53.8%). The mean change was equivalent to zero within ±2 mm (two one-sided tests p < 0.001).

Conclusions

Focal thoracic OPLL thickness remained predominantly stable after PDF, with threshold-level progression infrequent and neurological improvement maintained. Focal radiographic stability, rather than continued enlargement of the residual mass, was the predominant postoperative course after PDF.

Introduction

Ossification of the posterior longitudinal ligament (OPLL) produces ectopic bone that can compress the spinal cord and cause myelopathy.^1,2 Thoracic OPLL is less common than cervical OPLL but may cause severe deficits owing to the narrow thoracic canal, limited vascular reserve of the cord, and anterior spinal artery stenosis at sites of ossification.^3,4

The postoperative behavior of OPLL remains incompletely understood. ⁵ Traditionally viewed as progressive,^5,6 OPLL has more recently been shown in cervical studies to have its progression suppressed—and sometimes its thickness reduced—after posterior fusion versus motion-preserving procedures.^6-9

Thoracic OPLL is commonly treated with posterior decompression and fusion (PDF), which indirectly decompresses the cord while leaving the ossified mass in situ ¹⁰ ; however, longitudinal radiographic data on thoracic OPLL behavior after PDF remain limited.

We therefore characterized longitudinal CT-based changes in thoracic OPLL thickness after PDF, along with thoracic kyphosis, neurological outcomes, and exploratory correlates.

Methods

Study Design and Patient Selection

The study hypothesis was that posterior decompression and instrumented fusion (PDF) would be associated with predominantly stable focal thoracic OPLL thickness and limited progression, rather than continued enlargement of the ossified mass. This retrospective longitudinal study was conducted at a tertiary university hospital and is reported in accordance with the STROBE guideline for observational studies. Institutional review board approval was obtained; consent was waived (retrospective design).

We reviewed all consecutive adult patients (≥18 years) who underwent primary PDF for thoracic OPLL between 2009 and 2023. Inclusion criteria were CT-confirmed thoracic OPLL, ≥2 years of follow-up, and availability of preoperative and follow-up CT scans, standing radiographs, and clinical outcome data.

Exclusion criteria were revision surgery, concurrent cervical OPLL surgery, inadequate imaging, systemic bone-metabolic disease, and spinal malignancy or infection.

Variables included age, sex, body mass index (BMI), number of instrumented levels, and screw density (screws placed/available sites × 100%). Within each construct, vertebral levels were classified as instrumented (bilateral pedicle screws) or bridged (spanned without direct screw fixation).

Surgical Technique

All patients underwent PDF using a standardized technique. ¹¹ Pedicle screws were inserted and decompressive laminectomy with partial facetectomy or pediculectomy was performed as needed, followed by rod placement with attempted kyphosis correction. Screws were placed at every level when possible; when partial pediculectomy was required, screws were omitted at that vertebra, creating a bridged segment (no more than one bridged level per construct). Titanium alloy rods (6.0 mm) were used (Figure 1).

Figure 1.

Representative case showing reduced OPLL thickness after posterior fusion. Preoperative CT (a) sagittal and (b) axial views at the level of maximum compression; final follow-up CT (c) sagittal and (d) axial views at the same level. (e, f) Preoperative and postoperative lateral radiographs. Green calipers indicate OPLL thickness measurements

Radiographic Measurements

CT scans (2.5-mm slices) were obtained preoperatively and at follow-up, and reviewed on a picture archiving and communication system (Synapse 5; FUJIFILM, Tokyo, Japan) with an integrated digital caliper using bone-window settings. There was no fixed schedule for postoperative CT; in patients with more than one follow-up scan, the scan closest to the final clinical assessment provided the final measurement. No immediate postoperative CT was performed; the earliest postoperative imaging was a standing lateral radiograph obtained within approximately two weeks of surgery for sagittal alignment.

The level of maximal preoperative OPLL thickness was identified and tracked across time points. Maximum anteroposterior thickness was measured on axial images perpendicular to the posterior vertebral wall, excluding osteophytes (Figure 1); each measurement was repeated 3 times and averaged. Axial measurements served as the primary outcome, with sagittal measurements at the same level as complementary assessment.

Thoracic kyphosis (TK) was measured on standing lateral radiographs as the T5–T12 Cobb angle preoperatively, at the first postoperative standing radiograph, and at final follow-up, with a 5° threshold defining meaningful change.

OPLL morphology was classified as segmental, continuous, or mixed. ¹² The level of maximal ossification was categorized as upper (T1–4), middle (T5–8), or lower (T9–12) thoracic. Preoperative occupying ratio and lesion shape (beak, flat, or mixed) were recorded; morphology was reassessed at final follow-up.

Intra- and interobserver reliability, assessed in 15 cases (57.7%) by 2 independent observers (primary observer remeasuring after 2 weeks), was excellent (two-way random-effects intraclass correlation, ¹³ 0.928–0.997; Supplementary Table 1).

Outcome Definitions

The primary outcome was change in axial OPLL thickness from preoperative to final follow-up (negative values indicating reduction). A predefined 2-mm threshold classified patients as decreased, stable, or increased; this threshold was chosen to exceed measurement error and was prespecified as the minimal clinically important difference (MCID) for focal OPLL thickness, representing a conservative definition of meaningful change. ¹⁴ Directional decrease or increase was defined as any reduction or increase, respectively, regardless of magnitude.

Secondary outcomes included sagittal thickness change, TK, and neurological status by the modified Japanese Orthopaedic Association (mJOA) score for thoracic myelopathy, ¹⁵ recorded preoperatively, 10–14 days postoperatively, and at final follow-up. Recovery rate (RR) was calculated with the Hirabayashi formula [(final follow-up − preoperative mJOA)/(11 − preoperative mJOA)] × 100% ¹⁶ and graded excellent (≥75%), good (50–74%), fair (25–49%), or poor (<25%).

Statistical Analysis

Continuous variables are summarized as mean ± SD and categorical variables as counts (percentages); normality was assessed with the Shapiro–Wilk test. Paired comparisons used paired t-tests or Wilcoxon signed-rank tests, group comparisons used Mann–Whitney U, Kruskal–Wallis, chi-square, or Fisher exact tests, and correlations used Pearson or Spearman coefficients.

Repeated measures (TK, mJOA) used the Friedman test with Bonferroni-corrected post hoc Wilcoxon tests; the follow-up-duration trend in directional decrease used the Cochran–Armitage test (strata ≤3, 4–6, and >6 years, defined according to the postoperative year in which the final CT examination was obtained). To test the a priori hypothesis of stability rather than merely a non-significant difference, equivalence was assessed by two one-sided tests (TOST) on the primary axial change, with the predefined ±2-mm threshold as the equivalence margin. The threshold-level progression rate was compared with a literature-derived rate after motion-preserving decompression without fusion (54% at ≥2 years ¹⁴ ) by a one-sample exact binomial test. The multiple subgroup and correlate comparisons used the Benjamini–Hochberg false-discovery-rate procedure (hypothesis-generating). Sensitivity analyses used alternative thresholds (±1.0, ±1.5, ±3.0 mm), the interobserver minimal detectable change (MDC95), and outlier exclusion. Two-sided p < 0.05 was significant; analyses used Python 3.13 (SciPy, NumPy).

Results

Patient Characteristics

Of 38 patients screened, 12 were excluded (3 incomplete imaging, 5 follow-up <2 years, 2 concurrent cervical/lumbar surgery, 1 prior spinal surgery, 1 systemic bone disease), leaving 26 patients.

Baseline demographics are shown in Table 1. Most patients (82.6%) were overweight or obese; mean preoperative mJOA was 7.3 ± 1.8. Mean follow-up was 5.4 ± 2.7 years (range 2–11); 8 (30.8%) had ≤3, 9 (34.6%) had 4–6, and 9 (34.6%) had >6 years. Fourteen patients (53.8%) had a single follow-up CT and 12 (46.2%) had two or more. Maximal OPLL thickness was upper thoracic in 8 (30.8%), middle in 15 (57.7%), and lower in 3 (11.5%); baseline morphology was segmental in 14 (53.8%), continuous in 6 (23.1%), and mixed in 6 (23.1%). Mean instrumented levels were 9.7 ± 2.5 (range 5–14) (Table 2).

Table 1.

Patient Demographics and Clinical Outcomes

Variable	Total (N = 26)
Demographics
Age (years), mean ± SD	49.5 ± 12.4
Range	28–80
<50 years, n (%)	15 (57.7%)
50–64 years, n (%)	7 (26.9%)
≥65 years, n (%)	4 (15.4%)
Sex, n (%)
Male	15 (57.7%)
Female	11 (42.3%)
BMI (kg/m2), mean ± SD (n = 23)	32.7 ± 7.2
Range	21.3–47.5
Clinical Outcomes
mJOA, preoperative, mean ± SD	7.3 ± 1.8
mJOA, postoperative, mean ± SD	8.8 ± 1.9
mJOA, final follow-up, mean ± SD	8.7 ± 1.9
Recovery rate (%), mean	34.9
Outcome grade, n (%)
Excellent (RR ≥ 75%)	7 (26.9%)
Good (50–74%)	7 (26.9%)
Fair (25–49%)	1 (3.8%)
Poor (<25%)	10 (38.5%)
Not evaluable*	1 (3.8%)
Excellent/good combined	14 (53.8%)

Values are presented as mean ± SD or n (%). Recovery rate was calculated with the Hirabayashi formula. BMI data were available for 23 of 26 patients.

*Recovery rate could not be defined in 1 patient with a preoperative mJOA of 11 (Hirabayashi denominator of 0); this patient is shown separately and is not included in any outcome grade. BMI = body mass index; mJOA = modified Japanese Orthopaedic Association score; RR = recovery rate; SD = standard deviation.

Table 2.

Surgical Details

Variable	Total (N = 26)
Surgical
Instrumented levels, mean ± SD	9.7 ± 2.5
Range	5–14
Screw density (%), mean ± SD	80.5 ± 10.0
Range	54.5–100
Upper instrumented vertebra, n (%)
C7	3 (11.5%)
T1–T2	11 (42.3%)
T3–T5	10 (38.5%)
≥T6	2 (7.7%)
Lower instrumented vertebra, n (%)
T6–T10	6 (23.1%)
T11–T12	12 (46.2%)
L1–L3	8 (30.8%)

Values are presented as mean ± SD or n (%). LIV = lower instrumented vertebra; SD = standard deviation; UIV = upper instrumented vertebra.

Overall Thickness Changes

Mean axial OPLL thickness was 7.02 ± 1.58 mm preoperatively and 6.90 ± 2.01 mm at final follow-up (mean change −0.12 ± 1.03 mm). By the 2-mm threshold, 24 patients (92.3%) were stable, 1 (3.8%) decreased, and 1 (3.8%) increased (Figure 2); a directional decrease was observed in 18 (69.2%) and a directional increase in 8 (30.8%). A representative case is shown in Figure 1. The mean axial change was statistically equivalent to no change within the predefined ±2-mm margin (TOST p < 0.001; 90% CI of the mean change, −0.46 to 0.23 mm), with equivalence retained at margins down to ±0.5 mm (p = 0.036). The 3.8% threshold-level progression rate (1/26; 95% CI, 0.1–19.6%) was significantly lower than the 54% progression rate reported after motion-preserving cervical decompression without fusion (one-sample binomial p < 0.0001).

Figure 2.

Longitudinal OPLL thickness over follow-up. (a) Sagittal and (b) axial measurements (axial: primary outcome). Individual patient trajectories are colored by thoracic region; mean ± standard error is shown as black diamonds. Green shading: directional decrease (<0 mm); yellow shading: nonprogressive range (0 to <2 mm increase)

Sagittal measurements followed a similar pattern (mean change −0.18 ± 1.05 mm; directional decrease in 17 [65.4%], increase in 9 [34.6%]). Using the interobserver MDC95, 8 axial (30.8%) and 9 sagittal (34.6%) measurements met the threshold for detectable reduction, with 3 axial (11.5%) and 4 sagittal (15.4%) showing ≥1 mm decrease.

Exploratory Subgroup and Mechanical Analyses

Exploratory subgroup analyses are summarized in Table 3; none remained significant after correction for multiple comparisons. Directional decrease and mean axial change did not differ significantly by age, sex, body mass index (ρ = 0.027), baseline OPLL morphology, preoperative occupying ratio (ρ = 0.125), number of instrumented levels, or upper-instrumented-vertebra level (all p > 0.05). By region of maximal ossification, mean change was −0.58 mm (upper), +0.01 mm (middle), and +0.48 mm (lower thoracic), a nominal between-region difference that did not remain significant after correction for multiple comparisons (uncorrected p = 0.049; Benjamini–Hochberg q = 0.37). At final follow-up, morphology conversion occurred only from segmental to continuous (5 patients; 35.7% of baseline segmental lesions). Among 6 patients with OPLL spanning both instrumented and bridged segments, 5 (83.3%) showed greater reduction at instrumented segments (p = 0.313), and screw density was uncorrelated with change (ρ = 0.094); additional morphology and occupying-ratio data are in Supplementary Table 2.

Table 3.

Exploratory Subgroup Analyses of Axial OPLL Thickness Change

Subgroup	n	Any axial decrease, n (%)	p-value	Axial change, mean ± SD (mm)	p-value
By Age	​	​	1.000	​	0.304 (BH q = 0.49)
>50 years	10	7 (70.0%)	​	−0.49 ± 0.91	​
≤50 years	16	11 (68.8%)	​	+0.11 ± 1.06	​
By Sex	​	​	1.000	​	0.276 (BH q = 0.49)
Female	11	8 (72.7%)	​	−0.37 ± 1.07	​
Male	15	10 (66.7%)	​	+0.07 ± 1.00	​
By OPLL Type	​	​	0.691	​	0.296 a (BH q = 0.49)
Continuous	6	5 (83.3%)	​	−0.55 ± 0.54	​
Mixed	6	4 (66.7%)	​	−0.02 ± 1.02	​
Segmental	14	9 (64.3%)	​	+0.03 ± 1.19	​
By Region	​	​	0.211	​	0.049 a (BH q = 0.37)
Upper (T1–T4)	8	7 (87.5%)	​	−0.58 ± 0.45	​
Middle (T5–T8)	15	10 (66.7%)	​	+0.01 ± 1.23	​
Lower (T9–T12)	3	1 (33.3%)	​	+0.48 ± 0.71	​
By Fusion Length	​	​	1.000	​	0.533 b (BH q = 0.66)
>10 segments	11	8 (72.7%)	​	−0.23 ± 0.71	​
≤10 segments	15	10 (66.7%)	​	−0.04 ± 1.23	​
By Upper Fusion Level	​	​	0.401	​	0.068 b (BH q = 0.37)
T2 or above	14	11 (78.6%)	​	−0.47 ± 0.91	​
Below T2	12	7 (58.3%)	​	+0.29 ± 1.06	​

Overall cohort (N = 26): any axial decrease 18 (69.2%); mean axial change −0.12 ± 1.03 mm. Any axial decrease indicates axial thickness change < 0 mm at final follow-up (exploratory). Continuous axial change was the primary subgroup metric. Fisher exact test (binary proportions); chi-square test (3-group proportions).

^aKruskal–Wallis.

^bMann–Whitney U. Adjusted q values (Benjamini–Hochberg) computed across the full exploratory family (Supplementary Table 3); no comparison remained significant after adjustment. OPLL = ossification of the posterior longitudinal ligament.

Thoracic Kyphosis and OPLL Correlation

Mean TK increased by +6.2° ± 6.5° from preoperative to final follow-up (p < 0.001). TK change was not correlated with axial thickness change (r = 0.250, p = 0.218) or sagittal thickness change (r = 0.258, p = 0.204).

TK followed a 3-phase pattern (Friedman χ² = 15.680, p = 0.0004): 33.2° ± 10.6° preoperatively, 32.6° ± 10.5° at the first postoperative standing radiograph, and 39.4° ± 10.1° at final follow-up. Surgical correction was −0.6° ± 7.5° (p = 0.529) and follow-up loss was +6.8° ± 6.9° (p < 0.001); the increase occurred during follow-up rather than at surgery (Table 4).

Table 4.

Thoracic Kyphosis: Temporal Pattern and Correlation With Axial (Primary) and Sagittal OPLL Thickness Change

Variable	Value/vs Axial	Detail/vs Sagittal	P-value	Significance
A. Temporal Pattern (mean ± SD)
Preoperative TK	33.2 ± 10.6°	—	—	​
Immediate postoperative TK	32.6 ± 10.5°	Pre vs Post	0.529	NS
Final follow-up TK	39.4 ± 10.1°	Post vs Final	< 0.001	*
Friedman test (3 time-points)	χ 2 = 15.680	​	0.0004	*
B. TK Change Components	​	​	​	​
Surgical correction (Post − Pre)	−0.6 ± 7.5°	de-kyphosis ≥5° 26.9%; increase ≥5° 23.1%	0.529	NS
Correction loss (Final − Post)	+6.8 ± 6.9°	TK increase 80.8%; ≥5° 65.4%	< 0.001	*
Net TK change (Final − Pre)	+6.2 ± 6.5°	net increase ≥5° 65.4%	< 0.001	*
C. Correlation With OPLL Change	​	​	​	​
Surgical correction	r = +0.002 (p = 0.993)	r = +0.010 (p = 0.960)	​	NS
Correction loss	r = +0.233 (p = 0.252)	r = +0.231 (p = 0.256)	​	NS
Net TK change	r = +0.250 (p = 0.218)	r = +0.258 (p = 0.204)	​	NS

(A) TK (T5–T12) at 3 time points; Friedman test overall, Wilcoxon signed-rank with Bonferroni correction pairwise (adjusted α = 0.0167). (B) Decomposition of TK change; ≥5° considered to exceed routine measurement variability. (C) Pearson correlations between each TK component and OPLL thickness change. NS = not significant; OPLL = ossification of the posterior longitudinal ligament; TK = thoracic kyphosis. *P < 0.001. N = 26.

By the 5° threshold, 26.9% showed a decrease at the first postoperative standing radiograph, 50.0% were unchanged, and 23.1% showed an increase. During follow-up, 80.8% showed some TK increase and 65.4% an increase ≥5°, regardless of initial correction direction.

No TK component was correlated with axial or sagittal thickness change (Table 4 and Figure 3).

Figure 3.

Dissociation between thoracic kyphosis (TK) change and OPLL thickness change after posterior fusion. (a) Surgical correction. (b) Correction loss. (c) Net TK change. Axial (red triangles; primary outcome) and sagittal (blue circles) measurements. Dashed lines: linear regression; Pearson r and p values shown. No significant correlations (n = 26)

Follow-up Duration Analysis

Directional axial decrease increased with follow-up duration: 37.5% at ≤3 years (n = 8), 77.8% at 4–6 years (n = 9), and 88.9% at >6 years (n = 9) (Cochran–Armitage χ² = 5.136, p = 0.023; Figure 4).

Figure 4.

Relationship between follow-up duration and OPLL thickness change. (a, b) Scatter plots of sagittal (a) and axial (b) thickness change versus follow-up duration; dashed red lines: linear regression; Spearman ρ shown (†p < 0.10). (c, d) Proportions with directional decrease by time stratum for sagittal (c) and axial (d); Cochran-Armitage trend test (*p < 0.05; **p < 0.01)

Individual trajectories were plotted separately by direction (Figure 5). Serial measurements (≥2 postoperative CT) were available in 12 patients (7 decrease, 5 increase). In the decrease group, 5 declined steadily and the 2 with an early increase returned to or below baseline. In the increase group, none showed a monotonic increase: 3 peaked early and then declined—including the only patient exceeding +2 mm, from +3.10 to +2.96 mm—and 2 were essentially unchanged. In both groups peak thickness occurred early and was followed by stabilization or regression rather than sustained growth, with only 1 patient above the +2-mm threshold at final follow-up.

Figure 5.

Individual trajectories of OPLL thickness change by direction. Sagittal (a, b) and axial (c, d) change relative to the preoperative value, for the decrease (a, c) and increase (b, d) groups(groups defined by the direction of axial change, as in Table 5). Bold lines, patients with ≥2 postoperative CT (7 decrease, 5 increase); faint lines, single CT. Shaded band, ±interobserver minimal detectable change (0.44 mm axial, 0.42 mm sagittal); dashed lines, ±2-mm threshold

Sensitivity Analysis

Sensitivity analyses supported the descriptive pattern: after excluding 2 outliers the directional decrease rate was 70.8%, and no patient had an increase ≥3.0 mm (Supplementary Table 2).

Complications and Reoperations

No patient underwent reoperation at the index levels for OPLL progression, inadequate decompression, recurrent ventral compression, or construct failure. One patient (3.8%) required a series of early reoperations for deep surgical site infection with dural injury—wound closure (postoperative day 9), dural repair (day 13), extension of instrumentation to the ilium (day 14), irrigation and debridement (day 28), and staged rod removal (days 48 and 55)—after which all posterior instrumentation had been removed. No reoperation was related to progression of the ossified mass. Because posterior instrumented fixation was no longer maintained in this patient, a sensitivity analysis excluding this case yielded essentially unchanged results (axial change −0.17 ± 1.02 mm; directional decrease in 18 of 25 [72.0%]; equivalence within ±2 mm retained, TOST p < 0.001).

Clinical Outcomes

Mean mJOA improved from 7.3 ± 1.8 preoperatively to 8.8 ± 1.9 postoperatively and 8.7 ± 1.9 at final follow-up (Friedman χ² = 18.096, p = 0.0001; mean recovery rate 34.9%), with a combined excellent/good rate of 53.8% (14/26); the outcome-grade distribution is shown in Table 1.

Patients with an increase versus a decrease in thickness were compared (Table 5). Neurological outcomes were similar: mean mJOA improvement was +1.25 ± 2.14 (increase, n = 8) versus +1.47 ± 2.00 (decrease, n = 18) (p = 0.759), and excellent or good recovery occurred in 50.0% versus 55.6% (p = 1.000). Baseline severity, age, BMI, instrumented levels, screw density, OPLL type, and morphology did not differ. The increase group had a larger baseline ossified mass—greater preoperative axial thickness (8.29 ± 1.79 vs 6.46 ± 1.13 mm) and occupying ratio (median 60.0% vs 51.0%)—whereas the canal anteroposterior diameter was descriptively similar, indicating that the difference reflected the size of the ossified mass rather than canal narrowing. Because of the small sample and post hoc classification by the direction of measured change, these findings should be considered hypothesis-generating; consistent with this caution, the corresponding continuous analysis showed no association between preoperative occupying ratio and axial thickness change (ρ = 0.125, p = 0.542).

Table 5.

Comparison of Patients With an Increase Versus a Decrease in Axial OPLL Thickness at Final Follow-up

Variable	Increase group (n = 8)	Decrease group (n = 18)	P value
Neurological outcome
Preoperative mJOA	6.50 ± 1.51	7.69 ± 1.77	0.172
Final mJOA	7.75 ± 1.89	9.17 ± 1.72	0.081
mJOA improvement	+1.25 ± 2.14	+1.47 ± 2.00	0.759
Recovery rate, mean (%)	24.5 ± 44.9	39.9 ± 67.7	0.277
Recovery rate, median [IQR] (%)	28.6 [−15.3, 66.7]	66.7 [12.5, 100.0]	—
Excellent/good outcome, n (%)	4 (50.0)	10 (55.6)	1.000
Baseline and radiographic features
Age (years)	49.5 ± 17.0	49.5 ± 10.4	0.824
BMI (kg/m2)	32.6 ± 8.5	32.8 ± 6.8	0.922
Preoperative axial thickness (mm)	8.29 ± 1.79	6.46 ± 1.13	0.008
Preoperative occupying ratio, median (%)	60.0	51.0	0.024
Canal AP diameter (mm)	13.4 ± 0.4	12.9 ± 0.8	0.222
Instrumented levels (n)	9.9 ± 2.2	9.7 ± 2.7	1.000
Screw density (%)	79.1 ± 11.5	81.1 ± 9.6	0.802
Preoperative TK (°)	29.9 ± 9.6	34.7 ± 11.0	0.266
OPLL type (seg/cont/mixed)	5/1/2	9/5/4	0.691
Lesion morphology (beak/flat/mixed)	4/2/2	6/5/7	0.696
Region (upper/middle/lower)	1/5/2	7/10/1	0.211
Change exceeding MDC95 (0.44 mm), n (%)	5 (62.5)	8 (44.4)	—
Change ≥ 2 mm, n (%)	1 (12.5)	1 (5.6)	—

Groups were defined by the direction of axial thickness change from preoperative to final follow-up. Continuous variables are shown as mean ± SD unless otherwise stated and compared with the Mann–Whitney U test; categorical variables were compared with the Fisher exact or χ² test. Recovery rate could not be defined in 1 decrease-group patient with a preoperative mJOA of 11 (excluded from recovery-rate calculations). P values are unadjusted and derive from post hoc dichotomization; the comparisons are descriptive and not confirmatory, and the corresponding continuous analyses were not significant. AP, anteroposterior; BMI, body mass index; IQR, interquartile range; MDC₉₅, 95% minimal detectable change; mJOA, modified Japanese Orthopaedic Association; OPLL, ossification of the posterior longitudinal ligament; TK, thoracic kyphosis.

Discussion

Focal thoracic OPLL thickness after PDF remained stable in most patients over long-term follow-up. Directional decreases were frequent but typically small and below the threshold for measurable change, and neurological improvement was generally maintained. With a mean follow-up of 5.4 years (maximum 11), this represents, to our knowledge, one of the longest CT-based longitudinal characterizations of thoracic OPLL after PDF, supporting the view that posterior fusion may limit further enlargement.

Three features of this analysis strengthen the interpretation of stability. First, equivalence testing formally established that the mean thickness change fell within a predefined ±2-mm margin, so the finding reflects demonstrated stability rather than a non-significant test of difference in a small sample. Second, the 3.8% threshold-level progression rate was markedly lower than rates reported after motion-preserving decompression without fusion; because no comparable thoracic decompression-only series exists, this external, historical comparison with the cervical literature should be interpreted with caution. Finally, no exploratory subgroup or correlate remained significant after correction for multiple comparisons, indicating that overall stability, rather than any subgroup-specific effect, is the dominant signal.

The direction-stratified trajectories were also informative regarding the natural history of focal thickness. In both groups, peak thickness occurred early and was followed by stabilization or regression rather than continued growth: within the decrease group longer follow-up was accompanied by continued or stable reduction, and within the increase group the serially imaged patients either peaked early and then declined or remained essentially unchanged, with no patient showing sustained progression. This pattern indicates that the overall stability reflects a genuinely stabilizing process rather than the averaging of two divergent trajectories.

The surgical treatment of thoracic OPLL has shifted from decompression alone to posterior decompression and fusion,^17,18 with progression more often reported after decompression alone. ¹⁹ Case- and segment-based observations suggest that stabilization may also limit further ossification.^10,18,20-22 Our findings align with this framework: axial thickness remained predominantly stable and only 1 patient progressed.

Our thoracic findings parallel the cervical OPLL literature, where posterior instrumented fusion has been associated with suppressed progression and measurable thickness reduction versus motion-preserving procedures, particularly laminoplasty.^7-9,14 The thoracic pattern is similar in overall stability but smaller in magnitude—threshold-level decreases were uncommon—plausibly reflecting the lower baseline motion of the thoracic spine, which may reduce the dynamic stimulus for ossification and limit post-fusion remodeling. Because thoracic OPLL is not radiographically static without stabilization, ⁵ the observed stability is more plausibly attributable to surgical stabilization than to inherently static biology.

Directional decreases nonetheless predominated and increased with follow-up duration, possibly reflecting gradual remodeling; this should be interpreted cautiously given the small sample and non-threshold measures.

Cyclic mechanical stretch promotes osteogenic activity in OPLL-derived cells,^23,24 providing a plausible mechanism whereby fusion-induced reduction of motion could limit ossification, consistent with prior cervical and thoracic morphologic studies.^7,8,25 Metabolic factors were not evaluated. ²⁶

TK change was not associated with axial or sagittal OPLL thickness change, despite little immediate TK correction and progressive kyphosis during follow-up. Because TK was measured on standing radiographs and OPLL thickness on supine CT, these variables are complementary rather than directly interchangeable.

Most patients experienced meaningful neurological improvement. The lower recovery rate than in some prior reports^18,20,27,28 reflects a higher baseline mJOA: because the Hirabayashi denominator (11 − preoperative mJOA) limits achievable percentage improvement at higher baselines, the mean final mJOA remained favorable.

Because focal OPLL thickness remained stable in most patients with neurological improvement maintained long-term, PDF alone can be considered sufficient stand-alone treatment for most cases, without direct anterior resection of the ossified mass. When adequate indirect decompression cannot be achieved—typically with severe ventral cord compression—anterior or combined approaches should be considered at the index surgery. Posterior decompression for thoracic OPLL also carries a recognized risk of transient postoperative neurological deterioration,^29,30 underscoring the importance of achieving adequate but careful decompression. The decision thus hinges on the adequacy of indirect decompression rather than on concerns about progression of the residual mass.

Limitations include the retrospective single-center design, small sample size, heterogeneous follow-up, and absence of an internal comparison group, which limits causal inference. Bone metabolic markers and bone mineral density were not evaluated, and the within-patient comparison between instrumented and bridged segments was exploratory, involving only a small subgroup. These constraints bound the present conclusions to focal thickness behavior and preclude inferences about between-treatment causality. Nonetheless, the cohort’s size and follow-up duration are comparable with the largest published long-term series of thoracic OPLL after PDF, ²⁸ and this represents, to our knowledge, one of the few longitudinal CT-based characterizations of focal thickness changes in this population.

Conclusions

This study provides a long-term CT-based analysis of focal thoracic OPLL thickness after PDF. Most patients demonstrated radiographic stability by predefined thresholds, with small directional decreases predominating and becoming more common with longer follow-up, and neurological improvement was maintained. Focal radiographic stability, rather than continued enlargement of the residual ossified mass, was therefore the predominant postoperative course after posterior decompression and instrumented fusion.

Ethical Considerations

This retrospective study was approved by the Institutional Review Board of Hamamatsu University School of Medicine, Hamamatsu, Japan (approval no. 26-021), and was conducted in accordance with the 1964 Declaration of Helsinki and its later amendments.

ORCID iDs

Maiweilan Maihemuti https://orcid.org/0000-0002-5200-309X

Go Yoshida https://orcid.org/0000-0003-0265-6372

Tomohiro Banno https://orcid.org/0000-0002-6938-7501

Tomohiro Yamada https://orcid.org/0000-0002-7220-7321

Hideyuki Arima https://orcid.org/0000-0003-0586-6818

Hiroki Ushirozako https://orcid.org/0000-0002-7722-2205

Alex Quok An Teo https://orcid.org/0000-0002-0200-3331

Shin Oe https://orcid.org/0000-0002-0794-4026

Koichiro Ide https://orcid.org/0000-0003-4253-568X

Yu Yamato https://orcid.org/0000-0002-4701-8211

Consent to Participate

The requirement for written informed consent to participate was waived by the Institutional Review Board owing to the retrospective design; an opt-out method was used.

Data Availability Statement

The datasets generated and analyzed during the present study contain potentially identifying patient information and are therefore not publicly available. De-identified data may be available from the corresponding author on reasonable request, subject to institutional ethics committee approval and a data-sharing agreement.*